I need opinions, and hopefully suggestions for this mesh please!

[LisaMarie McWinnie]

Hello! So I've been working on an early 1900's dress, and I used for inspirations photos and drawings of dresses with the then-fashionable pidgeon bodice.

I am still using the mirror modifier, but still I am afraid the bodice still looks too mirroed, not natural. I accidentaly ended up making it looking more like pleates than what I actually wanted, but I asked someone IRL and they said it looks natural.

These are some of the inspirations:

http://www.kci.or.jp/archives/digital_archives/photos/113_xl_AC10092.jpg

http://www.kci.or.jp/archives/digital_archives/photos/105_xl_AC03638.jpg

http://24.media.tumblr.com/tumblr_lzepvx1nd11rnhcayo1_500.jpg

 

And now, the mesh:

I think it is not detailed enough as I would like, and I want more smaller detail after the bigger ones ends, similar to the first image. The sleeves are pretty much the same, I can't make them look natural!

 

Thank you in advance!

[Cathy Foil]

You are definitely not going to want to try and get all the fabric detail by creating more polygon geometry.  That would be just way too many vertices.

What you are going to need to use are the materials LL added a while back.  Since there isn't much reflections or shininess to the fabrics in your photos you will mainly need to master Normal Maps.  Normal Maps can give you all the detail you need.

Where ever the pattern of the fabric repeats be sure to put the UVs you make over one another basically stacking them.  This way one relatively small normal map texture can be repeated over your entire dress.  It is better to have a normal map texture of say 256x256 repeated 10 or 15 times over the dress than it is to have one 1024x1024 texture map covering the whole dress.  The 256x256 repeated will give you much more crisp and detailed result in SL. 256x256 multiplied by 10 would equal a texture that is 2560x2560 or if multiplied by 15 would be 3840x3840 and that is just one of the eight faces your mesh can have.  Repeating smaller textures in this way does not cause more lag it actually lessens it.

Each different pattern of fabric or normal map texture will count as one face even if the faces using that texture map in the dress are not touching one another.  You can use up to 8 faces per mesh.  Imagine a chessboard.  The white squares would be just one face and the black squares would be another face so the mesh would only contain two faces even if the corners of the squares were not touching.  By stacking the UVs of the white squares on top of each other you can use the same normal map for all of then.  Same goes for the black squares UVs.

Here are a couple of videos explaining Normal Maps in SL.

Here is a link to a program that can create simulated Normal Maps from 2D textures.

http://www.crazybump.com/

There is a free trial period.  It is not the best way to create normal maps but probably best in your case.

Hope that helps.
Cathy

[Dree Eames]

Mirroring geometry is often just the beginning for creating organic items that aren't supposted to be perfect.   After you apply the mirror modifier (in object mode) you take it a step further and move individual verts around, with or without proportional editing turned on, to create a more organic appearance that isn't all "matchy-matchy" from one side to the other.  You can also select groups of vertices and apply "smooth vertex' and relax them.  Or go into sculpting mode and soften/relax the pleates a bit with the smoothing brush.  (Take care to keep an original backed up while you play around).

You may also wish to use multiresolution modifier (which is similar to subsurf) and create a temporary high resolution version of your mesh into which you can use the sculpt function in blender to  sculpt finer detail and create a normal map full of details that you can apply as bump in SL materials to the lower poly base model.

[Chic Aeon]

I would like to echo Dree (hope I spelled that correctly ). Once you have the basics done with the mirror modifier THEN you can adjust to make things look less perfect and more realistic. I usually use proportional editing with the circle set quite small (I am in Blender so not sure how this translate ot other software) to push and pull groups of vertices.

And as for the normal maps, I agree that  you do NOT want more geometry. I have a fair amount of items (usually hair and shoe and not dresses thank goodness) that disappear each time I teleport because there are WAY too many triangles in their making. Normal maps will give you definition. BUT REMEMBER THAT NOT THAT MANY PEOPLE CAN SEE THEM  -- at the moment anyway. Firestorm hasn't released its out of beta materials viewer and since the beta has issues ....

And even when most folks are on material viewers, they may not be able to set their graphics to see normal maps so beting totally dependent on them is not a good idea. AND normal maps are directional by light source.   I DO use them (other grid now and some here when they were new) but sparingly. So along WITH the normal maps -- there is nothing better then great texturing!

 

My two cents .

[Dree Eames]

I agree Chic.  You really want to put your effort in the tweaking and the texturing.  Normal maps should perhaps be viewed as the frosting on the cake for those who can view them. The mesh should stand on it's own as if they weren't added, but if you can squeeze in a bit more detail, it might be worth adding them.  I've been told all these extra images do add lag so that might be an argument against using them.  I'm not inworld much these days so I have little personal experience to use to make that judgement call.  

Here is a very ugly example snapshot from SL.  I made by quickly modeling a very basic test mesh to which I added multi res and sculpted (please don't throw rotten tomatoes at me -  I spent a whole 5 minutes modeling this thing )

Both shirts are the same low poly (1 LI) object:

 

[Madeliefste Oh]

Your sleeves look good, once textured it will have a natural look.

I think the pleats in your top are too much point shaped. The fabric in your reference pictures seems thin, like fine woven cotton.However the pleats you have look give the impression that the fabric is very thick, like heavy canvas or velvet maybe. Looking at the picture the pleats seem more broad and more flat.

It's also a bit more bloused at the belly. It would look natural when the fabric falls party over the belt.

 

 

[Dree Eames]

I haven't done any clothes rigging since mesh was in Beta, and even then I was a beginner at it, but the first thing that strikes me is the arms  being modeled in the "T" orientation. (unless you've pre-rigged this and you are displaying it with the arms down)   I'm making an assumption that you are going to want to wear this item (although it might be a shop window display).

I wonder if anyone here who rigs clothing can make a recommendation for this op about this.

[Coby Foden]

---

Dree Eames wrote:

... but the first thing that strikes me is the arms  being modeled in the "T" orientation.

---

I have always thought this arms modelling in T-pose being very strange too. Wouldn't A-pose be more natural? Where does this T-pose modelling originate from I wonder?

Jonathan Williamson, in his female figure modelling series, used A-pose. The reason being that he saw A-pose more natural pose than T-pose to model a human figure. (T-pose raises the shoulders up in a position where we very seldom see them.)

[Helium Loon]

---

Coby Foden wrote:

I have always thought this arms modelling in T-pose being very strange too. Wouldn't A-pose be more natural? Where does this T-pose modelling originate from I wonder?

Jonathan Williamson, in his female figure modelling series, used A-pose. The reason being that he saw A-pose more natural pose than T-pose to model a human figure. (T-pose raises the shoulders up in a position where we very seldom see them.)

---

The primary reason for the T-Pose is that it aligns the major extremities along the XYZ axes.  This makes for simpler storage and calculations of deformations and rotations from the base (axis-aligned) orientation.  This was from back when such calculations could take hours or days to do.  It's kind of hung around since then.  And since most animation is done using skinned vertex shaders these days, it makes a simpler shader program if you know that the bones will be basically aligned with a specific axis....

 

[Cathy Foil]

---

Coby Foden wrote:

I have always thought this arms modelling in T-pose being very strange too. Wouldn't A-pose be more natural? Where does this T-pose modelling originate from I wonder?

Jonathan Williamson, in his female figure modelling series, used A-pose. The reason being that he saw A-pose more natural pose than T-pose to model a human figure. (T-pose raises the shoulders up in a position where we very seldom see them.)

---

Coby you can model and rig your mesh with you skeleton in an A-pose and then just export your mesh while the skeleton is in the T-pose.

Here's how I do it in Maya.  First rig a separate mesh to the skeleton while it is in T-pose it can be something as simple as a cube or a single face.  This becomes the Bind Pose for the skeleton.  I am sure there is another way of doing this but this is quick and simple.  Then hide the simple mesh.  Then rotate the joints till you get your skeleton in the A-pose you want or any other pose such as a four legged animal pose.  Then model your mesh to fit this new pose.  Next skin and weight the mesh while the skeleton is still in the A-pose.  Then in the Skin menu click Go To Bind Pose.  Since your mesh is already weighted to the skeleton the mesh will move into the T-pose ready for export.

I hope you find that helpful.

Cathy

[LisaMarie McWinnie]

Thank you all for the replies!

I have used normal maps in some of my last creations, and I like it alot, though for now I don't want to rely so much in it. Same thing with the specularity, is not as strong in world as i would like.

I agree Madelle, that is probably one of the things that is bugging me right now, it is looking thick while I want it to appear like a thin delicate fabric.

And Dree, I always model my clothes with the arms open, in the neutral pose, for this picture I put the arms down.

[Coby Foden]

---

Cathy Foil wrote:

 

Coby you can model and rig your mesh with you skeleton in an A-pose and then just export your mesh while the skeleton is in the T-pose.

---

Thanks Cathy. :smileyhappy:  Now I just need to learn how do it in Blender as I don't use Maya.

[LisaMarie McWinnie]

Oh now I have a new questions:

Can I put the arms of the SL Avatar down on Blender, like in my picture, model the mesh with the avatar like that, and make it rigged? I tried this before but it didn't worked well.

[Cathy Foil]

---

Coby Foden wrote:

Thanks Cathy. :smileyhappy:  Now I just need to learn how do it in Blender as I don't use Maya.

---

There is another way in Maya that doesn't use the Go To Bind Pose command and I am sure it would work for Blender as well.

You just basically have to animate the skeleton in a few frames.  Frame one have the skeleton in the standard T-pose.  Then in frame two or three put the skeleton in the A-pose and key frame it.  With the animation still in frame two or three create your mesh to fit the A-pose and rig it and adjust the weights.  Then rewind back to frame one and your skeleton and mesh will be in the T-pose ready for export.  I think that should work for Blender as well. 

Hope that helps.

Cathy

[Medhue Simoni]

---

Coby Foden wrote:

---

Dree Eames wrote:

... but the first thing that strikes me is the arms  being modeled in the "T" orientation.

---

I have always thought this arms modelling in T-pose being very strange too. Wouldn't A-pose be more natural? Where does this T-pose modelling originate from I wonder?

---

I would think it comes from wanting the skeleton's bones to all start from a position with no rotation. All motion capture systems have a reference frame using a Tframe. This is done to match up the bones to the points, or the shape of the reference figure.